Functional and clinical neuroanatomy of morality

(1)

BRAIN

A JOURNAL OF NEUROLOGY

REVIEW ARTICLE

Functional and clinical neuroanatomy of morality

Manuela Fumagalli and Alberto Priori

Dipartimento di Scienze Neurologiche, Universita` degli Studi di Milano, Fondazione IRCCS Ca’ Granda, Ospedale Maggiore Policlinico, 20122 Milano, Italy

Correspondence to: Prof. Alberto Priori, MD, PhD Dipartimento di Scienze Neurologiche,

Fondazione IRCCS Ca’ Granda, Ospedale Maggiore Policlinico, Via Francesco Sforza 35, 20122 Milano, Italy E-mail: alberto.priori@unimi.it

Morality is among the most sophisticated features of human judgement, behaviour and, ultimately, mind. An individual who behaves immorally may violate ethical rules and civil rights, and may threaten others’ individual liberty, sometimes becoming violent and aggressive. In recent years, neuroscience has shown a growing interest in human morality, and has advanced our understanding of the cognitive and emotional processes involved in moral decisions, their anatomical substrates and the neurology of abnormal moral behaviour. In this article, we review research findings that have provided a key insight into the functional and clinical neuroanatomy of the brain areas involved in normal and abnormal moral behaviour. The ‘moral brain’

consists of a large functional network including both cortical and subcortical anatomical structures. Because morality is a complex process, some of these brain structures share their neural circuits with those controlling other behavioural processes, such as emotions and theory of mind. Among the anatomical structures implicated in morality are the frontal, temporal and cingulate cortices. The prefrontal cortex regulates activity in subcortical emotional centres, planning and supervising moral decisions, and when its functionality is altered may lead to impulsive aggression. The temporal lobe is involved in theory of mind and its dysfunction is often implicated in violent psychopathy. The cingulate cortex mediates the conflict between the emotional and the rational components of moral reasoning. Other important structures contributing to moral behaviour include the subcortical nuclei such as the amygdala, hippocampus and basal ganglia. Brain areas participating in moral processing can be influenced also by genetic, endocrine and environmental factors. Hormones can modulate moral behaviour through their effects on the brain. Finally, genetic polymorphisms can predispose to aggressivity and violence, arguing for a genetic-based predisposition to morality. Because abnormal moral behaviour can arise from both functional and structural brain abnormalities that should be diagnosed and treated, the neurology of moral behaviour has potential implications for clinical practice and raises ethical concerns. Last, since research has developed several neuromodulation techniques to improve brain dysfunction (deep brain stimulation, transcranial magnetic stimulation and transcranial direct current stimulation), knowing more about the ‘moral brain’ might help to develop novel therapeutic strategies for neurologically based abnormal moral behaviour.

Keywords:aggressiveness; behavioural neurology; behaviour; brain; psychiatry

Abbreviations:COMT = catechol-O-methyltransferase; tDCS = transcranial direct current stimulation; MAOA = monoamine oxidase A

doi:10.1093/brain/awr334 Brain 2012: Page 1 of 16 | 1

Received February 10, 2011. Revised September 16, 2011. Accepted October 25, 2011.

For Permissions, please email: journals.permissions@oup.com

by guest on February 27, 2014http://brain.oxfordjournals.org/Downloaded from

(2)

Introduction

Abnormal moral behaviour implies the violation of ethical rules and civil rights and often results in violence and criminal acts.

Moral behaviour is the product of a complex process that although somehow primed by genes and environment, is ultimately con- trolled by the brain. Since the 1990s, both behavioural and neu- roscientific studies of morality have grown in volume and sophistication. In particular, the interest in the neurobiological de- terminants of human ethics prompted neuroscientists to address several intriguing issues, including whether specific neuronal net- works or anatomical brain structures exist for moral behaviour.

Besides bringing about a major advance in neuroscience and in the biology of mind, understanding the neural foundations of morality could help in developing novel strategies for treating abnormal moral behaviours. It also has potentially important social implications. For instance, developing a treatment for serial sexual assaulters could help prevent further crimes, reduce the number of victims and diminish the need for imprisonment and ultimately, benefit society.

In this article, to understand better the functional and clinical neuroanatomy of morality, we review and integrate findings from studies in various fields (clinical neurology and psychiatry, neuroimaging, neurophysiology, neuropathology, behavioural genetics and psychoneuroendocrinology) investigating moral judgement, moral behaviour and violent behaviour, considered as the behavioural expression of moral abnormality. Unlike previous reviews on morality (Moll et al., 2005; Raine and Yang, 2006; Young and Koenigs, 2007;Funk and Gazzaniga, 2009;Huebner et al., 2009;

Moll and Schulkin, 2009), we discuss studies on healthy subjects and on patients with neurological and psychiatric disorders through a neuroanatomical approach. To highlight the complexity and multiple faceted neural control of morality we also consider the role of genes and hormones. Although abnormal moral behaviour can be part of several neurological and psychiatric disturbances, for the purposes of this review we focus on conditions primarily characterized by violent behaviour that could be attributed to moral abnormality.

Violence as the behavioural consequence of moral

abnormality

Several tests can now be used to measure morality in experimental studies (tasks that use philosophical moral dilemmas, tasks that present visual sentences and tasks that present pictures) (Greene et al., 2001;Moll et al., 2001;Harenski and Hamann, 2006), and for clinical purposes, scales and questionnaires are available to assess abnormal moral behaviour (Rush et al., 2008).

From a behavioural point of view, the major consequence of moral abnormality is violence. Violence stems from multiple factors including social learning, frustration, cognitive and moral development, mental and neurological disorders, environmental and genetic factors (Fabian, 2010). Their cumulative effect portends

a behaviourally negative outcome, such as psychopathy, criminality, murder, substance abuse disorder, homelessness and un- employment (Fabian, 2010). Aggression can be defined as ‘any form of behaviour directed towards the goal of harming or injuring another living being who is motivated to avoid such treatment’

(Baron and Richardson, 2004). Aggression can be secondary to focal cerebral disease or related to brain abnormalities in individuals with no other overt signs of a cerebral disorder (David et al., 2009). Aggression becomes violence when it is exaggerated and unjustifiable, leading to violation of social norms and to destruc- tion (Victoroff, 2009). To fulfil the aims of this review, among psychiatric disorders we will consider psychopathy, sociopathy and antisocial personality disorders. A critically important denom- inator shared by all these conditions is the failure to conform to the commonly accepted mores of society that often ultimately results in abnormal moral behaviour.

Frontal lobe

Healthy subjects

Neuroimaging studies demonstrate that a brain region closely involved in moral judgement is the ventromedial prefrontal cortex (Greene et al., 2001; Moll et al., 2001, 2002a, b;

Heekeren et al., 2003;Borg et al., 2006;Harenski and Hamann, 2006;Luo et al., 2006;Prehn et al., 2008;Harenski et al., 2009;

Young and Saxe, 2009; Shenhav and Greene, 2010; Sommer et al., 2010). The ventromedial prefrontal cortex has a critical role in encoding the emotional value of sensory stimuli (Rolls, 2000), and its recruitment during moral decisions supports ventromedial prefrontal cortex involvement in their emotional processing (Greene et al., 2001;Moll et al., 2002a,b;Heekeren et al., 2003).

The ventromedial prefrontal cortex is also implicated in allowing one to adhere to social norms and cultural values (Moll et al., 2005), and in representing preferences about others’ outcomes according to perceived intentions (Cooper et al., 2010). The left ventromedial prefrontal cortex is more intensely activated in subjects with lower ability to apply moral principles in justifying the resolution of moral dilemmas (Prehn et al., 2008). Ventromedial prefrontal cortex/medial orbitofrontal cortex activation also correl- ates with the ‘expected moral value’ of decision options, defined as the expected number of lives lost/saved (Shenhav and Greene, 2010). Promising research developments over recent years include studies designed to modulate moral judgement with non-invasive brain stimulation techniques [transcranial direct current stimulation (tDCS) and transcranial magnetic stimulation] over the different areas of the frontal lobe. tDCS entails applying weak direct currents through an electrode on the scalp for a time ranging from seconds to minutes (Fox, 2011). This technique induces pro- longed changes in cortical excitability: anodal stimulation generally increases and cathodal stimulation decreases cortical excitability (Priori, 2003;Utz et al., 2010), respectively depolarizing or hyper- polarizing the neuronal membrane (Accornero et al., 2007).

When Koenigs et al. (2009) delivered bilateral prefrontal tDCS to 25 subjects doing a moral judgement task they failed to obtain statistically significant results. In a study conducted in our

(3)

laboratory (Fumagalli et al., 2010), we administered a moral judgement task to 78 healthy subjects before and after anodal or cathodal tDCS applied over the ventral prefrontal cortex or occipital cortex. In female subjects, we found that, whereas anodal prefrontal tDCS significantly increased utilitarian responses, cathodal prefrontal tDCS tended to decrease them. According to Greene et al. (2008) and Greene (2007), a utilitarian choice aims at ‘maximizing benefits and minimizing costs across affected individuals’ (Greene et al., 2008), ‘endorsing harmful actions that promote the greater good’ (Greene, 2007). Hence, anodal ventral prefrontal tDCS interferes with utilitarian decisions specifically in females. We hypothesized that anodal stimulation reduces the prefrontal activity in emotional evaluation and does so by mimick- ing the effect of a lesion. A possible explanation for gender difference is that tDCS modulates the female tendency towards altruism more easily than the male tendency. Whereas altruism in males ‘resists’ tDCS-induced changes, altruism in females might be more sensitive to small, transient changes induced by stimulation.

Because dopamine is an anionic catecholamine that migrates towards the anode during electrophoresis, anodal ventral prefrontal tDCS might act by increasing dopamine levels in the frontal lobe, thus influencing the dopamine reward circuit and ultimately altering decisional processes, increasing utilitarian response rates.

Though several methodological reasons make it difficult to compare the study conducted by Koenigs et al. (2009) with the one conducted in our laboratory (Fumagalli et al., 2010), our finding provides new evidence that brain stimulation (tDCS) is able to modulate moral judgement and behaviour (at least under laboratory conditions).

The rational counterpart of the ventromedial prefrontal cortex is the dorsolateral prefrontal cortex. This frontal lobe area is involved in problem-solving and cognitive control (Greene et al., 2004) and competes with the ventromedial prefrontal cortex in suppressing prepotent emotional reactions (Greene et al., 2004). Specifically, in moral judgement the dorsolateral prefrontal cortex has a pivotal role in aggregating cost-benefit analysis and in utilitarian moral reasoning (Greene et al., 2001, 2004), intervenes in evaluating situations that recruit rule-based knowledge (Prehn et al., 2008), and in deciding on responsibility for crimes and on appropriate punishment (Haushofer and Fehr, 2008). The dorsolateral prefrontal cortex is also activated during dishonest behaviour, probably reflecting the process of lying (Greene and Paxton, 2009). tDCS over the dorsolateral prefrontal cortex also influences deceptive responses (Priori et al., 2008;Mameli et al., 2010).

Other brain regions involved in moral processing are the cingulate cortex (Greene et al., 2001,2004;Harenski et al., 2008) and the left subgenual gyrus (Luo et al., 2006). The anterior cingulate cortex mediates the ‘conflict’ between the emotional and the rational components of moral reasoning, whereas the posterior cingulate cortex is more closely related to emotion and social ability (Greene et al., 2001,2004).

In conclusion, data in healthy subjects show that the frontal lobe drives moral behaviour and that the role of the various frontal areas probably differs during moral decision processing: whereas the orbital and ventromedial prefrontal cortices emotionally drive moral decisions, the dorsolateral prefrontal cortex acts mainly as a

rational ‘filter’. This dual opposite processing involving the ventromedial prefrontal cortex and the dorsolateral prefrontal cortex seems to be mediated by the anterior cingulate cortex.

Patients

The first clinical evidence showing a link between personality, behaviour, morality and the frontal lobe came from studies on patients. In 1848, while Phineas Gage was working on a railroad line, an accidental explosion drove a steel rod up into his left cheek and through his frontal lobes. A local physician, Harlow, described him as fitful, irreverent, impatient of restraint or advice when it con- flicts with his desires, and also obstinate and capricious (Feldman and Goodrich, 2001). In 1936, on the basis of animal experiments conducted by John Fulton, Egas Moniz proposed surgically cutting the connections between the frontal cortex and the thalamus (leucotomy) to treat certain psychoses (Heller et al., 2006). In his article on psychosurgery, Moniz reported good results also in severely agitated, anxious or depressed patients (Moniz, 1937).

Quenching the original enthusiasm for brain surgery as a viable means of combating mental illness, in 1942 Walter Freeman reported the results for 200 lobotomy cases and acknowledged that the procedure was not always benign: 14% of the patients who underwent prefrontal lobectomy subsequently manifested seizures and impaired cognition, affect, mood and social behaviour (Feldman and Goodrich, 2001). After surgery these patients also exhibited a severe frontal syndrome, altered abstract thinking and judgement (Feldman and Goodrich, 2001), and various personality changes. Freeman described patients as

‘. . . lazy, rude, boisterous, restless and inane [. . .] [the patient] is relatively unteachable, having lost those social skills that are necessary for living outside an institution [. . .]. If the patient has previously demonstrated antisocial traits such as alcoholism, drug addiction, criminality, avoidance of responsibility, aggressiveness or psychopathic activities, the effect of operation may be to free him from any residual sense of guilt or shame, and thus turn loose upon society an individual whose behavior is intolerable’ (Freeman, 1950).

Hence, surgical procedures involving the frontal cortex and frontal tracts reportedly achieved successes or failures: the symptoms were reduced or, conversely, as Freeman reported, exacer- bated. The main reason for these opposing results is probably the lack of homogeneity in the procedures used by Moniz and Freeman (Freeman, 1950; Knight, 1969). The lack of neuroimaging techniques, the psychopathological heterogeneity of patients, the lack of quantitative outcome variables and of neuropathologic- al controls in studies reporting the results of frontal leucotomy and other psychosurgical interventions make it difficult to understand from historical data how surgery affects the frontal lobe in humans. Whatever the clinical results, surgery and lesions of the frontal lobe can induce moral changes thus further demonstrating the key role of this structure in moral behaviour.

Reviewing 25 previously published studies of post-lesion acquired antisocial personality disorders and 39 studies of post-lesion acquired obsessive–compulsive disorders, Braun et al. (2008)

(4)

found that the most typical lesion causing antisocial personality disorder is that involving the ventromedial prefrontal cortex, occasionally spreading to the caudate nucleus (Braun et al., 2008). In an early study by Feuchtwanger (1923) cited inDavid et al. (2009),comparing patients with frontal gunshot wounds with patients in whom bullets penetrated other parts of the skull, frontal lesions induced euphoria, irritability, aggressivity, apathy, attention deficits and moral defects (David et al., 2009). Acquired lesions to the frontopolar and to the ventromedial prefrontal cortex at an early age may lead to even more severe impairments in moral behaviour, suggesting that the prefrontal cortex is involved in moral development (Eslinger et al., 1992;Anderson et al., 1999). Patients with ventromedial prefrontal cortex lesions also have a significantly high frequency of moral disturbances with aggressive and violent behaviour.

They also typically seem disinhibited, impulsive and unconcerned with the consequences of their behaviour (Rolls et al., 1994;

Grafman et al., 1996;Brower and Price, 2001), a neurological syndrome significantly defined as ‘acquired sociopathy’ (Saver and Damasio, 1991). These observations are consistent with animal studies showing that lesions in the orbitofrontal cortex are related to uncontrolled aggression (Butter et al., 1970;Elliott, 1990). Patients with ventromedial prefrontal cortex damage retain the knowledge of social rules, recognize the elements that compose social situations, for example current contingencies, possible response options and future outcomes (Saver and Damasio, 1991), but they cannot make appropriate decisions in natural settings. This inadequate decisional processing makes them insensitive to the future positive or negative consequences of an action. The ‘somatic marker model’

explained the reduced anticipatory skin conductance in response to stimuli predicting negative outcomes observed in these patients as an inability to mark the implications of a decision with a signal that automatically distinguishes advantageous from pernicious actions (Damasio et al., 1990;Damasio, 1994). Systematic studies about the impairment of social and moral behaviour related to damage to the ventromedial prefrontal cortex reported various explan- ations. For example they proposed that moral judgements in these patients are characterized by an abnormal high rate of utilitarian judgements for moral dilemmas in which social emotions play a pivotal role in resolving moral conflict (Ciaramelli et al., 2007;Koenigs et al., 2007;Moretto et al., 2009;Thomas et al., 2011) or by more morally permissive judgements for failed attempts to harm (Young et al., 2010b). Focusing on how patients justify moral judgements, others suggest that patients’

justifications do not reflect normal adult levels of moral reasoning but are characterized by egocentrism (Anderson et al., 1999).

The ‘somatic marker model’ and these data suggest that moral reasoning relies on both rational/conscious and emotional/uncon- scious processing in the ventromedial prefrontal cortex. Prefrontal involvement, measured in terms of reduced grey matter in this area, is also reported in antisocial and psychopathic individuals (Raine et al., 2000;Yang et al., 2005), and in aggressive patients with temporal lobe epilepsy (Woermann et al., 2000). In the neuroimaging literature, reduced prefrontal glucose metabolism is associated with acts of aggressive impulsive behaviour (Goyer et al., 1994), with antisocial personality disorder (Bufkin and Luttrell, 2005), and characterizes murderers compared with

normal controls (Raine et al., 1994). Superior and ventrolateral prefrontal activity also positively correlated with psychopathy in an emotion regulation task (Harenski et al., 2009), whereas in moral/non-moral picture distinctions, ventromedial prefrontal cortex activity is more severely reduced in psychopaths than in non-psychopaths (Harenski et al., 2010) (Fig. 1). New imaging evidence for the neural bases of moral and prosocial sentiments comes from a study in patients with the behavioural variant of frontotemporal dementia engaged in a moral sentiment task that highlighted a role of the medial frontopolar cortex (Moll et al., 2011).

Anterior cingulate cortex lesions result in persistent behavioural changes, in particular reduced and occasionally increased aggressivity, disinhibition, impulsivity, emotional blunting and decreased motivation (Devinsky et al., 1995).

Current knowledge from studies in patients therefore suggests that the frontal structures most directly involved in abnormal moral behaviour are the medial orbitofrontal cortex and ventromedial prefrontal cortex.

Temporal lobe and insula

Healthy subjects

A further important structure for moral behaviour is the temporal lobe. Neuroimaging studies found that a temporal lobe region involved in moral judgement is the superior temporal sulcus (Moll et al., 2001, 2002a, b; Heekeren et al., 2003, 2005;

Greene et al., 2004; Borg et al., 2006; Harenski and Hamann, 2006). Superior temporal sulcus activates during the elaboration of moral dilemmas because it is associated with emotion (Greene et al., 2004), during processing of social cognition mechanisms (Moll et al., 2002b;Greene et al., 2004) and in making decisions about complex ethical dilemmas (Heekeren et al., 2003; Greene et al., 2004).

Another region implicated in moral function includes the anterior/middle temporal gyrus (Moll et al., 2001; Heekeren et al., 2003, 2005;Greene et al., 2004; Harenski and Hamann, 2006;

Sommer et al., 2010). Accordingly, some researchers retain that these areas are activated after the point of decision, in particular when subjects choose a utilitarian response, suggesting that neural processing of moral conflict is associated with higher cognitive demands (Greene et al., 2004;Sommer et al., 2010).

A further temporal region activated during moral dilemma evaluation (Greene et al., 2001) and involved in evaluating moral agency and responsibility (Borg et al., 2006) is the angular gyrus.

The temporo-parietal junction (Young et al., 2007;Kedia et al., 2008;Harenski et al., 2009;Young and Saxe, 2009) contributes to moral intuition (spontaneous, unsolicited attention directed towards cues that have potential moral salience such as a person in distress, weapons or emotion-laden words) (Harenski et al., 2009), and to belief attribution during moral judgement (Young et al., 2007;Young and Dungan, 2011). Neurostimulation studies also support a direct link from theory of mind to moral judgement.

Theory of mind is a person’s intuitive ability to understand other

(5)

people’s plans, thoughts, points of view, beliefs, attitudes and emotions. Transcranial magnetic stimulation applied to the right temporo-parietal junction before or during a moral judgement task led subjects to rely less on the actor’s mental states, judging attempted harms as less morally forbidden and more morally per- missible (Young et al., 2010a).

Finally, insular cortex activation is found during various moral tasks (Moll et al., 2002b;Greene et al., 2004;Borg et al., 2008;

Kedia et al., 2008;Cooper et al., 2010) and is related to emotional processing (Greene et al., 2004; Kedia et al., 2008), to disgust processing (Moll et al., 2002b) and to detecting and processing uncertainty (Cooper et al., 2010). Other studies reported that the insula is more strictly involved in moral processing and sensitive to norm violations implicated in deontological judgement (Huebner et al., 2009):it is activated in the perception of inequity (Hsu et al., 2008) and in care and justice cognition (Caceda et al., 2011).

In conclusion, available findings in healthy subjects imply that, although not necessarily engaged in moral decisions, the insular cortex is involved in emotional processing and in encoding inequity. The anterior/middle temporal gyrus has a secondary role, being recruited during working memory and general cognitive processing. Whereas the superior temporal sulcus

evaluates the intentionality and social complexity of moral actions, the temporo-parietal junction has a role in belief attribu- tions and theory of mind, two basic requirements for moral processing.

Patients

The first observation about the temporal lobe’s role in moral and violent behaviour comes from psychosurgery. In 1891, Burckhardt removed the temporal lobe in five patients with intractable psychiatric disturbances who he described as demented and aggressive, with unsuccessful results (Berrios, 1997). Temporal lobe resection is a surgical procedure also used to treat patients with temporal lobe epilepsy, who often experience psychiatric, cognitive and behavioural manifestations (Hamberger and Drake, 2006;Schramm, 2008;David et al., 2009). This procedure is able to solve behavioural disorders, although patients continue to experience cognitive disturbances involving memory and language.

Although no studies have specifically investigated moral decisions in patients with temporal lobe epilepsy and in patients who underwent temporal lobe resection, their behavioural abnormalities could at least partly reflect damage to limbic structures, in particular the amygdala. Anatomical studies showed reduced temporal Figure 1 Analyses of brain activity during moral picture viewing showed a significant interaction between psychopaths/non-psychopaths and moral versus non-moral and neutral picture viewing in the anterior temporal cortex (A) and in ventromedial prefrontal cortex (B). This finding indicates that non-psychopaths showed a significantly greater moral than non-moral and neutral picture distinction in these regions, whereas psychopaths did not. Error bars are standard errors. (Copyright ß 2010 by the American Psychological Association.

Reproduced with permission fromHarenski et al., 2010. The use of APA information does not imply endorsement by the APA).

(6)

lobe volume in patients with early-onset conduct disorder (Kruesi et al., 2004),in incarcerated psychopaths (Dolan et al., 2002) and in antisocial personality disorders (Barkataki et al., 2006). In line with anatomical observations, neuroimaging studies documented reduced temporal lobe activation in aggressive patients (Volkow and Tancredi, 1987;Amen et al., 1996), psychopaths (Soderstrom et al., 2000) and violent offenders (Raine et al., 2001). Electro- encephalographic recordings also disclose right temporal lobe abnormalities in patients with aggressive antisocial behaviour (David et al., 2009).

Structural abnormalities and volume losses in the right superior temporal gyrus have been reported in psychopathic individuals (Mu¨ller et al., 2007 cited in Weber et al., 2008). Activation deficits in this brain area are reported in antisocial and psychopathic patients during a semantic processing task, showing that psychopaths used more cognitive resources to process affective information than did healthy subjects (Kiehl et al., 2004).

Reduced bilateral activation in the middle temporal gyrus is also a typical neuroimaging finding in aggressive children with temporal lobe epilepsy (Juhasz et al., 2001), antisocial patients (Goethals et al., 2005), violent psychiatric patients (Volkow et al., 1995) and violent offenders (Seidenwurm et al., 1997;

Soderstrom et al., 2000).

Reduced activation is reported also in the anterior temporal cortex in violent patients (Wong et al., 1997) and in psychopaths during a moral decision-making task (Fig. 1) (Harenski et al., 2010). Anterior temporal lobe involvement in psychopaths seems to be related specifically to reduced ability to process morally sa- lient stimuli, restricting moral processing to generation of semantic and emotional context (Harenski et al., 2010).

Among temporal structures, the hippocampus has an important role in emotion and behaviour. A significant negative correlation was found between the posterior hippocampal volume and degree of psychopathology in repetitive violent offenders (Laakso et al., 2001), and a specific abnormal hippocampal morphology in the absence of total grey matter volume changes characterizes habit- ually violent offenders (Boccardi et al., 2010). These data fit well with the hippocampal role in the acquisition and retrieval of fear conditioning (Burman et al., 2006;Tsetsenis et al., 2007), and in aggressive and impulsive behaviour (Guillot et al., 1994; Sluyter et al., 1996;Deakin, 2003;Prior et al., 2004;Cardinal, 2006;van Goozen and Fairchild, 2006). Last, hippocampal stimulation in humans elicits severe rage reactions (Heath, 1992). Using depth electrodes in patients with epilepsy with aggressive behaviour, Saint-Hilaire et al. (1981) recorded epileptic activity localized to the right hippocampus during a spontaneous aggression.

Finally, ictal electroencephalographic recordings in aggressive patients with temporal seizures show that the epileptic activity during spontaneous epileptic aggressive behaviour localizes first in the right amygdala, then in the right temporal cortex, right hippocampus and parahippocampal gyrus and then reaches the anterior and median cingulate gyrus and right supplementary motor area (Saint-Hilaire et al., 1981).

Right insular lesions are associated with low emotional intelli- gence, poor judgement in decision-making and disturbances in social functioning (Bar-On et al., 2003). Bilateral grey matter reduction in the insula are reported in psychopaths

(de Oliveira-Souza et al., 2008). Fear conditioning activates insula in healthy subjects but not in psychopaths (Birbaumer et al., 2005), suggesting that the emotional processing for antici- pated pain and for threat stimuli is impaired in patients.

In conclusion, studies from patients argue that various temporal areas, in particular the superior temporal gyrus, the middle temporal gyrus and the anterior temporal cortex, are closely related to moral judgements and violent behaviour. The hippocampus is consistently associated with violent–aggressive behaviour and is involved in the development of moral constitution (Laakso et al., 2001).

Parietal lobe

Healthy subjects

Some functional neuroimaging studies report inferior parietal lobe activation during moral processing (Greene et al., 2004;Harenski et al., 2008;Caceda et al., 2011). The study by Greene et al.

(2004) suggests that this brain area is associated with working memory and cognitive control, so that it is recruited because the task proposed engaged cognitive processing.

Patients

Whereas, to our knowledge, no published data are available on parietal lesions or abnormalities and moral behaviour, neuroimaging studies disclose increased blood flow in the parietal cortex in non-psychotic violent offenders (Soderstrom et al., 2000). In murderers (Raine et al., 1997) and individuals with impulsive personality disorders (Siever et al., 1999) metabolism—investigated using functional MRI and PET—was reduced in the superior parietal cortex.

In conclusion, despite the scant literature, although available data argue that the parietal lobe contributes to moral judgement and is involved in abnormal moral behaviour, its precise function in processing moral decisions is still unknown.

Subcortical structures

Healthy subjects

Only few functional MRI studies have found subcortical involvement during moral processing. The amygdala activates during processing of both basic and moral emotions (Moll et al., 2002b), during evaluation of moral judgement (Greene et al., 2004), during violation of severity ratings for moral pictures (Harenski et al., 2008) and during personal desire-oriented decisions in contrast to morally guided responses (Sommer et al., 2010).

Amygdala activation is reduced during passive viewing of moral and non-moral pictures when individuals attempted to decrease emotions (Harenski and Hamann, 2006) and in response to bodily harm (Heekeren et al., 2005). Amygdala activation induced by bodily harm may reflect amygdala-specific involvement in affective judgements about the emotional value of an action

(7)

(Heekeren et al., 2005), and could imply reduced moral brakes when the individual is in danger.

Bilateral thalamic activation can be observed during decisions on whether to follow a moral rule or fulfil a personal desire (Sommer et al., 2010), whereas activation in the septal region is related to charitable donations (Moll et al., 2006).

The caudate nucleus is activated during the evaluation of moral stimuli (Luo et al., 2006), in charitable living (Harbaugh et al., 2007) and in altruistic punishment (de Quervain et al., 2004).

The amygdala therefore appears to be a pivotal subcortical structure involved in processing moral emotions.

Patients

Evidence that subcortical structures intervene in morality is further supported by data from patients. Disinhibition and violent behaviour are related to grey matter loss or volume loss or structural amygdala abnormalities in violent offenders (Wong et al., 1997;

Tiihonen et al., 2000;Van Elst et al., 2000) and in psychopathic individuals (Yang et al., 2006). Why amygdalotomy improves 30–

40% of patients with severe behavioural disturbances, aggressiveness and violent behaviour (Fountas and Smith, 2007) therefore remains unclear. A possible explanation is that amygdala dysfunction causes patients’ behavioural abnormalities and aggressiveness.

In these instances, surgery to remove dysfunctional amygdala can restore normal behaviour. Neuroimaging studies during various tasks report inconsistent results on amygdala responses: two functional MRI studies reported increased amygdala activation in antisocial individuals viewing negative visual content (Mu¨ller et al., 2003) and during an aversive conditioning task (Schneider et al., 2000). In contrast, others report reduced amygdala activation during the processing of negative affective stimuli in criminal psychopaths (Kiehl et al., 2001), in fear conditioning in criminal psychopaths (Veit et al., 2002;Birbaumer et al., 2005), in response to emotional stimuli in adolescents with conduct disorders (Sterzer et al., 2005) and during emotional moral decision-making in psychopathy (Glenn et al., 2009). Another study comparing psychopaths and healthy subjects during a moral decision-making task found that psychopaths lacked a positive association between amygdala activity and the severity of moral violation ratings, suggesting a reduced emotional response (Harenski et al., 2010).

Collectively, findings in patients agree with those in healthy subjects and confirm a pivotal role of the amygdala in processing moral emotions. In line with structural and neuroimaging studies, stimulation studies also showed amygdala involvement in emotion and violence expression.Heath et al. (1955)described the case of a patient receiving amygdala stimulation who became enraged and attacked the psychiatrist. Some studies in Treiman’s (1991) review also reported aggressive behaviour during amygdala stimulation. The amygdala might therefore represent a sort of ‘switch’

between brain processing of moral emotions and their translation into action and aggression.

Impulsive-antisocial temperament of psychopathic patients is related to neurochemical and functional nucleus accumbens alterations during monetary reward anticipation. The role of this structure in psychopathic behaviour probably reflects its participation

in the mesolimbic dopaminergic system, involved in reward and impulsive behaviour (Buckholtz et al., 2010).

Early evidence suggesting the role of the septum in empathy, guilt and remorse in psychopathy (Gorenstein and Newman, 1980) has now been corroborated by a PET study reporting hypometabolism in the septal region while patients with the behavioural variant of frontotemporal dementia executed a moral sentiment task (Moll et al., 2011).

Accidental stimulation of the posteromedial hypothalamic area by deep brain stimulation induced aggressive behaviour in patients with Parkinson’s disease (Bejjani et al., 2002;Sensi et al., 2004). In contrast, deep brain stimulation of the posterior hypothalamus improved patients with disruptive behaviour and mental retard- ation (Franzini et al., 2005; Hernando et al., 2008;Kuhn et al., 2008).

In patients with Parkinson’s disease, subthalamic nucleus deep brain stimulation can occasionally induce mild personality changes, disinhibition, mood changes, hypersexuality and patho- logical gambling (Temel et al., 2005), ultimately influencing moral behaviour. Neurophysiological studies over the past 10 years have described subthalamic activity changes in local field potentials recorded through the stimulation electrodes im- planted for deep brain stimulation in patients with Parkinson’s disease engaged in various cognitive and behavioural tasks (Kuhn et al., 2005; Brucke et al., 2007;Marceglia et al., 2009, 2011; Fumagalli et al., 2011). In a study conducted in our laboratory to assess the role of basal ganglia in moral decision- making and in the processing of moral conflictual stimuli (Fumagalli et al., 2011), we recorded subthalamic local field potentials from 16 Parkinsonian patients doing a moral task. The task required patients to evaluate moral conflictual, moral non-conflictual and neutral sentences, and to respond if they agreed or disagreed by pressing one of two buttons. The results showed that a specific subthalamic rhythm, low-frequency band (5–13 Hz), is involved in decision processing: the subthalamic nucleus intervenes in moral judgement in a conflict- dependent way, its oscillations being specifically modulated by moral conflictual stimuli. These findings extend the ‘moral brain’

to encompass the basal ganglia including the subthalamic nucleus, especially for conflict processing (Fumagalli et al., 2011) (Fig. 2).

In summary, available data argue for subcortical participation in the brain network responsible for moral reasoning. The amygdala has a crucial role in processing social and emotional content and in particular in learning that some actions harm others and should therefore be avoided (Blair, 2007). Amygdala structural and functional impairments mean that psychopathic individuals and crim- inals are unable to recognize cues that signal threat, becoming relatively fearless and more readily engaged in antisocial behaviour (Sterzer, 2010). Finally, the septal area is involved in empathy and altruism and its hypometabolism is associated with impairments on prosocial sentiments, such as pity and guilt (Moll et al., 2011).

Emerging evidence suggesting that subcortical structures intervene in abnormal moral behaviour might indicate directions for future research on the therapeutic effects of deep brain stimulation in psychopathic disorder and violent behaviour.

(8)

Genes and hormones

Two centuries ago, the Italian criminologist Cesare Lombroso suggested that phenotypic features revealed the criminal attitude of the individual. The general concept that genetic alterations, abnormalities in the brain and its functioning can be linked to criminal behaviour nowadays attracts interest (Pietrini and Bambini, 2009). Geneticists have shown that complex behavioural disorders can partly arise from variations in DNA sequences and environmental factors. Family, twin and adoption studies have all suggested that genetic factors contribute to antisocial personality disorders and externalizing behaviours. Genetic and non-shared environmental factors are both important in the development of

psychopathy. The weight of the two factors is debated, but increasing evidence shows that genetic factors account for 49–81% of the variance and in psychopathy are more important than environment (Gunter et al., 2010).

The brain catecholaminergic systems regulate aggressive behaviour. Two major enzymes are responsible for catecholamine catabolism in the brain: monoamine oxidase A (MAOA) and catechol-O-methyltransferase (COMT). The MAOA gene is located on the X chromosome. The MAOA enzyme preferentially catabo- lizes the brain neurotransmitters noradrenaline and serotonin (Volavka et al., 2004). The remarkable consequences of MAOA gene deletions in mice suggest that mild functional variations in this gene probably have significant behavioural impacts Figure 2 Subthalamic local field potentials in patients with Parkinson’s disease during a moral decision task. (A) Time–frequency plots of grand average subthalamic nucleus local field potentials in each task condition (moral conflictual, moral non-conflictual, neutral). The x-axis shows the time (s) from the mean reaction time value of the task condition to 1 s after the choice. The sentence evaluation phase starts from sentence presentation to 500 ms before key pressing (time = 0). Note that during the moral task, low-frequency power (5-13 Hz) increased during sentence evaluation. (B) Low-frequency (LF) power modulation during the sentence evaluation for each task condition shows higher modulation during moral conflictual sentences. (C) Histograms represent the mean low-frequency modulation in sentence evaluation during each task condition. Note the significantly higher power modulation during the moral conflictual condition. In conclusion, subthalamic local field potential recordings show strong subthalamic nucleus activation especially during conflictual moral decisions. Error bars represent the confidence interval of the estimated mean (1.96 standard error). (Reprinted fromFumagalli et al., 2011, with kind permission of Taylor & Francis Ltd. http://www.informaworld.com).

(9)

(Shih et al., 1999). In humans, the first proof of a link between aggressivity and MAOA came from a study investigating a Dutch family with a repeated incidence of violent criminal behaviour among males across several generations all of whom had a MAOA gene abnormality (Brunner et al., 1993). Low MAOA activity, combined with traumatic early life events, seems to increase the likelihood of antisocial disturbances in adults (Caspi et al., 2002;

Meyer-Lindenberg et al., 2006;McDermott et al., 2009) (Fig. 3).

Low enzymatic COMT activity is associated with increased brain catecholamine levels that raised the likelihood of aggression (Volavka et al., 2004). Consistent with this expectation, COMT knockout mice exhibited increased aggressive behaviour (Gogos et al., 1998) and in humans, the COMT Met allele is associated with aggressive behaviour in males (Craig and Halton, 2009).

As well as genetic factors, hormones can profoundly influence behaviour and cognition by interfering with brain metabolism and neuronal function. Areas implicated in social and emotional processing have extensive connections with peripheral physiologic- al functioning, and specifically, have strong reciprocal connections

with the endocrine system. Whereas the hypothalamus–pituitary–

gonadal axis is involved in the reproductive and immune systems (Terburg et al., 2009), the hypothalamic–pituitary–adrenal axis is entailed in the limbic emotion-related neurocircuitry. The end product of the hypothalamic–pituitary–adrenal axis is cortisol and the end product of the hypothalamus–pituitary–gonadal axis is testosterone. Cortisol is a key modulator of several emotion- related learning and memory processes, of social behaviour, and is also associated with anxiety symptoms. The brain network involved in cortisol regulation includes the prefrontal cortex, insula, amygdala and anterior cingulate cortex. In particular, the prefrontal cortex is implicated in hypothalamic–pituitary–adrenal axis inhibition and regulation and insular activity in response to traumatic stimuli is associated with adrenocorticotropic hormone responsivity (Shirtcliff et al., 2009). Finally, anterior cingulate cortex hypoactivation indicates reduced cortisol reactivity and is associated with callousness (Shirtcliff et al., 2009), whereas hyper- activation means increased cortisol reactivity and is related to social stressors (Eisenberger et al., 2007). Extensive behavioural A 102

B

100 98 96 94 92 90 88

120

110

100

90

Low High

Sex Region Left amygdala Right amygdala Supragenual cingulate Subgenual cingulate

Males Females MAOA genotype

Percentage volume relative to MAOA-H mean ± 1 SEMPercentage volume relative to male MAOA-H mean ± 1 SEM

Low High

MAOA genotype

Figure 3 The correlation between the MAOA polymorphism and brain structure and function assessed with MRI in healthy humans.

(A) Individuals with the low-expression variant associated with an increased risk of violent behaviour compared with subjects with the high-expression variant exhibit significant volume reductions in bilateral amygdala, supragenual anterior cingulate and subgenual anterior cingulate cortex. (B) Male individuals with the low-expression variant show greater lateral orbitofrontal volume, bilaterally, than subjects with the high-expression variant. No genotype effect was found in females, indicating a highly significant sex genotype interaction.

Plots represent the summed volumes of voxels in predefined region of interest, normalized to volume measures relative to the high- expression variant of the MAOA gene (MAOA-H) group mean (100%). [Reprinted fromMeyer-Lindenberg et al., 2006, with kind permission of National Academy of Sciences, USA—Copyright (2011) National Academy of Sciences, USA].

(10)

endocrinology research associated reduced hypothalamic–pituitary–adrenal axis functioning in children and adolescents with the development of psychopathy (van Honk and Schutter, 2006).

Lower cortisol levels characterize psychopathic patients (Holi et al., 2006;O’Leary et al., 2007;Cima et al., 2008), patients with antisocial disorder with limited empathy and guilt and con- stricted emotionality (Hawes et al., 2009), and adolescents with conduct disorder (Cappadocia et al., 2009;Victoroff, 2009).

Another hormone involved in aggressive behaviour is testosterone. High testosterone levels enhance attention to aggressive stimuli, downregulate the interaction between cognitive and emotional brain systems and are associated with dominant aggressive behaviour (Dabbs et al., 1995;Dabbs and Hargrove, 1997).

During moral decision-making, individuals having high testosterone levels are more likely to make utilitarian decisions, especially when doing so implies acts of aggression and social cost (Carney and Mason, 2010). Testosterone is also associated with diminished sensitivity to the affective signals that facilitate pursuit of empathic behaviours and choices (van Honk et al., 2004, 2005, 2010).

Although the gender factor is largely ignored in neuroimaging and lesional studies on morality, testosterone might partially ex- plain the gender differences highlighted by previous behavioural studies (Gilligan, 1982;Jaffee and Hyde, 2000).

Finally, because a heightened testosterone level alone is often not enough to induce violence (Terburg et al., 2009), the hormonal marker for aggressive behaviour is the testosterone/cortisol ratio. When high testosterone is combined with low cortisol,

aggression is unmanageable and the subject could become a danger to society (Terburg et al., 2009;van Honk et al., 2010).

We found no studies investigating the correlation between the performances in a moral task with either COMT/MAOA genetic polymorphism or testosterone/cortisol levels in healthy subjects, although this could be an interesting issue for future research.

Another possible shortcoming worth investigating is that these studies measured basal hormone concentrations, without consider- ing receptor site functioning, possible interactions with other hormones and the mechanisms that regulate protein transcription and hormone secretion, all factors that can ultimately influence the behavioural effects of a given hormone, independently from its concentrations (Carney and Mason, 2010).

An overall view

As well as depending on the various underlying cortical areas, moral functions and abnormal moral behaviour depend on a distributed functional neural network connecting cortical and subcortical structures (Fig. 4andTable 1), modulated by neurotransmitters and hormonal systems. Recent behavioural evidence showing abnormal moral reasoning in patients with callosotomy also suggest that this neural network is not confined to one side of the brain (Miller et al., 2010). Because morality is a complex process, some of the brain structures involved share their neural circuits with those controlling other behavioural processes, such as

Figure 4 A schematic diagram reporting the hypothetical role of various brain areas in regulating moral behaviour.

(11)

Table 1 List of studies subdivided according to brain area and type of study

Brain structure Evidence from References

Frontal lobe

Frontopolar cortex Lesional studies Eslinger et al., 1992;Anderson et al., 1999 Orbitofrontal cortex—

ventromedial prefrontal cortex

Studies in healthy subjects Rolls, 2000; Greene et al., 2001; Moll et al., 2001,2002a,b,2005; Heekeren et al., 2003;Borg et al., 2006;Harenski and Hamann, 2006;Luo et al., 2006;

Prehn et al., 2008;Harenski et al., 2009;Young and Saxe, 2009;Cooper et al., 2010;Fumagalli et al., 2010;Shenhav and Greene, 2010;Sommer et al., 2010 Psychopathy and antisocial

behaviour

Goyer et al., 1994;Raine et al., 1994,2000;Bufkin and Luttrell, 2005;Yang et al., 2005;Harenski et al., 2009,2010

Lesional studies Damasio et al., 1990;Saver and Damasio, 1991;Eslinger et al., 1992;Rolls et al., 1994; Grafman et al., 1996;Anderson et al., 1999; Brower and Price, 2001;

Ciaramelli et al., 2007;Koenigs et al., 2007;Braun et al., 2008; Feuchtwanger, 1923 (cited inDavid et al., 2009);Moretto et al., 2009;Thomas et al., 2011;

Young et al., 2010b Frontotemporal dementia Moll et al., 2011

Epilepsy Woermann et al., 2000

Dorsolateral prefrontal cortex Studies in healthy subjects Greene et al., 2001, 2004;Haushofer and Fehr, 2008;Prehn et al., 2008;Greene and Paxton, 2009

Cingulate cortex Studies in healthy subjects Greene et al., 2001,2004;Harenski et al., 2008 Temporal lobe

Temporal lobe Lesional studies David et al., 2009

Psychopathy and antisocial behaviour

Volkow and Tancredi, 1987; Amen et al., 1996;Soderstrom et al., 2000;Raine et al., 2001;Dolan et al., 2002;Kruesi et al., 2004;Barkataki et al., 2006;David et al., 2009

Epilepsy Saint-Hilaire et al., 1981

Anterior/middle temporal lobe Studies in healthy subjects Moll et al., 2001;Heekeren et al., 2003,2005;Greene et al., 2004;Harenski and Hamann, 2006;Sommer et al., 2010

Volkow et al., 1995;Seidenwurm et al., 1997; Wong et al., 1997;Soderstrom et al., 2000;Goethals et al., 2005;Harenski et al., 2010

Epilepsy Juhasz et al., 2001

Superior temporal sulcus Studies in healthy subjects Moll et al., 2001,2002a,b;Heekeren et al., 2003,2005;Greene et al., 2004;Borg et al., 2006;Harenski and Hamann, 2006

Superior temporal gyrus Psychopathy and antisocial behaviour

Kiehl et al., 2004; Mu¨ller et al., 2007 (cited inWeber et al., 2008)

Temporo-parietal junction Studies in healthy subjects Young et al., 2007,2010a;Kedia et al., 2008;Harenski et al., 2009;Young and Saxe, 2009;Young and Dungan, 2011

Angular gyrus Studies in healthy subjects Greene et al., 2001;Borg et al., 2006

Insula Studies in healthy subjects Moll et al., 2002b;Greene et al., 2004;Borg et al., 2008;Hsu et al., 2008;Kedia et al., 2008;Cooper et al., 2010;Caceda et al., 2011

Birbaumer et al., 2005;de Oliveira-Souza et al., 2008 Hippocampus Psychopathy and antisocial

behaviour

Laakso et al., 2001;Boccardi et al., 2010 Epilepsy Saint-Hilaire et al., 1981

Parietal lobe

Parietal cortex Psychopathy and antisocial behaviour

Soderstrom et al., 2000

Inferior parietal lobe Studies in healthy subjects Greene et al., 2001,2004;Harenski et al., 2008;Caceda et al., 2011 Superior parietal cortex Psychopathy and antisocial

behaviour

Raine et al., 1997;Siever et al., 1999 Subcortical structures

Amygdala Studies in healthy subjects Moll et al., 2002b; Greene et al., 2004; Heekeren et al., 2005; Harenski and Hamann, 2006;Harenski et al., 2008;Sommer et al., 2010

Wong et al., 1997;Schneider et al., 2000;Tiihonen et al., 2000;Van Elst et al., 2000;Kiehl et al., 2001;Veit et al., 2002;Mu¨ller et al., 2003;Birbaumer et al., 2005;Sterzer et al., 2005;Yang et al., 2006;Fountas and Smith, 2007;Glenn et al., 2009;Harenski et al., 2010

Stimulation Heath et al., 1955;Treiman, 1991 Septum Studies in healthy subjects Moll et al., 2006

Frontotemporal dementia Moll et al., 2011 Psychopathy and antisocial

behaviour

Gorenstein and Newman, 1980

Caudate nucleus Studies in healthy subjects de Quervain et al., 2004;Luo et al., 2006;Harbaugh et al., 2007 Nucleus accumbens Psychopathy and antisocial

behaviour

Buckholtz et al., 2010 Subthalamic nucleus Deep brain stimulation studies Fumagalli et al., 2011 Thalamus Studies in healthy subjects Sommer et al., 2010

Posterior hypothalamus Deep brain stimulation studies Bejjani et al., 2002;Sensi et al., 2004;Franzini et al., 2005;Hernando et al., 2008;

Kuhn et al., 2008